Score:0

AWS cdk environment strategy: one per branch, one per developer, or only dev/staging/prod?

tt flag

we're currently setup with a dev, staging, and prod environments on aws. We're finding it hard to do QA properly, since often many commits happen in a short span of time, and are all included in a codepipeline build, which makes it hard to associate failures to a specific commit.

We were looking into spinning up one environment per feature branch, in a manner similar what this aws quickstart example is doing:

enter image description here

However, I'm finding it hard to justify spinning up our entire backend (which is huge) to test, in some cases, a single apigateway->lambda->dynamodb route. Furthermore, this can work for serverless services, but we also use elasticsearch server. In such a case, it doesn't even seem possible to spin an ES server just for testing a new feature branch. But if we point our feature branch to, say, staging's ES server, how do we make sure not to pollute it in case of bugs?

How do people usually go about solving this problem?

Oscar De León avatar
la flag
Pasting the chart from the vanilla quickstart example doesn't help to understand *your* problem. Are you using the same pipeline for everything, or do you have separate pipelines for each CodeCommit branch? can't you do specific testing without the need to spin up a full-sized backend?
samlaf avatar
tt flag
Right now we have 10ish pipelines (for each service) for 3 branches (so 30ish in total): dev, staging, and prod. What do you mean by specific testing?
mangohost

Post an answer

Most people don’t grasp that asking a lot of questions unlocks learning and improves interpersonal bonding. In Alison’s studies, for example, though people could accurately recall how many questions had been asked in their conversations, they didn’t intuit the link between questions and liking. Across four studies, in which participants were engaged in conversations themselves or read transcripts of others’ conversations, people tended not to realize that question asking would influence—or had influenced—the level of amity between the conversationalists.